Resilience, Migration, and Business Continuity · Topic 3 of 6
Disaster Recovery Strategies and Continuous Replication
Match AWS disaster recovery to RTO, RPO and cost: backup and restore, pilot light, warm standby, active active, and Elastic Disaster Recovery replication.
- 5 min
- RTO is downtime; RPO is data loss. Both drive the strategy.
- Cost rises from backup and restore, to pilot light, to warm standby, to active/active.
- Pilot light keeps data live but compute off; warm standby keeps a small fleet running.
What it is
Disaster recovery (DR) is how a workload comes back after an event that takes out a site, a Region or the data itself. The business sets two targets: the recovery time objective (RTO), the longest acceptable outage, and the recovery point objective (RPO), the most data loss it can accept, measured in time.
Lower RTO and RPO cost more, because more of the environment must already run in the recovery location. The exam gives the targets and a budget and asks for the cheapest strategy that still meets both.
How it works
The four DR strategies
| Strategy | What runs in the recovery Region | Typical RTO and RPO | Cost |
|---|---|---|---|
| Backup and restore | Only backups and copies | Hours | Lowest |
| Pilot light | Data replicated, core services off or minimal | Tens of minutes | Low |
| Warm standby | A scaled down but working copy | Minutes | Higher |
| Multi-site active/active | Full capacity serving traffic | Near zero | Highest |
Backup and restore relies on copies of backups in another Region, plus infrastructure as code to rebuild the stack. Pilot light keeps data live in the recovery Region, for example a cross-Region read replica, and starts compute only during recovery.
Warm standby already runs a smaller fleet, so recovery is mostly scaling up and switching traffic. Active/active serves users from both Regions all the time, which needs data that can be written in more than one Region.
Data replication choices
The data tier usually decides the achievable RPO.
| Data store | Cross Region option | Note |
|---|---|---|
| Amazon Aurora | Aurora Global Database | Storage based replication with low lag; promote a secondary Region in a failover |
| Amazon RDS | Cross Region read replica | Promote the replica during recovery |
| Amazon DynamoDB | Global tables | Multi Region, multi active writes |
| Amazon S3 | Cross Region Replication | S3 Replication Time Control adds a replication time SLA |
| Amazon EBS and others | Snapshot copies with AWS Backup | RPO equals the backup frequency |
Backups alone can never give an RPO shorter than the time between backups. When the question demands an RPO of seconds, look for continuous replication.
AWS Elastic Disaster Recovery
AWS Elastic Disaster Recovery (AWS DRS) replicates on-premises or cloud-based servers into a staging area subnet in a chosen AWS Region. The staging area uses low-cost storage and minimal compute, so you do not pay for a full standby fleet.
During an event you launch recovery instances within minutes, using the latest state or an earlier point in time. Point-in-time recovery matters for ransomware or corruption, because you can launch from a moment before the damage.
AWS DRS converts servers to boot natively on EC2, supports non-disruptive drills, and can replicate back to the primary site for failback when the issue is resolved.
| Feature | AWS Transform MGN | AWS Elastic Disaster Recovery |
|---|---|---|
| Purpose | Move servers to AWS once | Keep servers recoverable on AWS |
| Replication | Continuous until cutover | Continuous for as long as protection is needed |
| End state | Source retired | Failback to source possible |
| Point in time recovery | Not the goal | Launch from an earlier point in time |
Sizing the strategy to the requirement
A strategy that over-delivers wastes money. A batch reporting system with a 24-hour RTO does not need warm standby; backup and restore with copies in another Region is enough.
A server-based application with an RPO of seconds and an RTO of minutes, but no budget for a second running fleet, is the classic AWS DRS case. Replication is continuous, yet compute is only paid for at full size after a launch.
How the exam asks about this
- "RPO of seconds and RTO of minutes for 200 on-premises servers, at the lowest cost." -> AWS Elastic Disaster Recovery with a staging area in AWS.
- "RTO of 24 hours, minimal cost, data must survive a Region outage." -> Backup and restore with cross-Region backup copies and infrastructure as code.
- "Database data must be current in a second Region, but application servers can start during the event." -> Pilot light with Aurora Global Database or a cross-Region read replica.
What to remember for the exam
- RTO is downtime; RPO is data loss. Both drive the strategy.
- Cost rises from backup and restore, to pilot light, to warm standby, to active/active.
- Pilot light keeps data live but compute off; warm standby keeps a small fleet running.
- Backups cannot beat their own schedule for RPO; continuous replication can.
- AWS DRS gives seconds of RPO and minutes of RTO with low-cost staging resources.
- Point-in-time launches recover from corruption that has already replicated.
Replication copies mistakes too. A deleted table or encrypted files replicate to the DR site within seconds, so pair replication with point-in-time recovery or immutable backups.
Common mistakes
Choosing active/active by default. It meets every target but costs the most. If the stated RTO is hours, a cheaper strategy is correct.
Confusing pilot light and warm standby. Pilot light has no serving fleet running; warm standby already runs a smaller fleet that can take traffic at once.
Using AWS Transform MGN for DR. MGN is built for a one-time migration and cutover. Ongoing protection with drills and failback is AWS Elastic Disaster Recovery.
Check yourself
What separates pilot light from warm standby?
Pilot light replicates data but keeps compute off or minimal, while warm standby runs a scaled down working copy that only needs to scale up.
Why is AWS Elastic Disaster Recovery cheaper than a second running fleet?
It replicates into a staging area with low-cost storage and minimal compute, and launches full-size instances only for drills or recovery.
Why do teams need point-in-time recovery when they already replicate continuously?
Continuous replication copies corruption or ransomware damage, so recovery must start from a moment before the damage.
Topic 3 of 6 in Resilience, Migration, and Business Continuity
Test yourself on Resilience, Migration, and Business Continuity
The full AWS SAP-C03 question bank has 850 questions, with timed practice exams and an explanation for every answer.
This study guide is independent and is not affiliated with, endorsed by, or sponsored by AWS. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.